Analyzing Nature's Protective Design the Glyptodont Body Armor

Journal of the Mechanical Behavior of Biomedical Materials 82 (2018) 218–223

Contents lists available at ScienceDirect Journal of the Mechanical Behavior of Biomedical Materials journal homepage: www.elsevier.com/locate/jmbbm

Analyzing nature's protective design: The glyptodont body armor T ⁎ Anton du Plessisa, , Chris Broeckhovenb, Igor Yadroitsevc, Ina Yadroitsavac, Stephan Gerhard le Rouxa a CT Scanner Facility, Central Analytical Facilities, Stellenbosch University, Matieland, 7602 Stellenbosch, South Africa b Laboratory of Functional Morphology, Department of Biology, University of Antwerp, Wilrijk 2610, Belgium c Department of Mechanical and Mechatronic Engineering, Central University of Technology, Bloemfontein 9300, South Africa

ARTICLE INFO ABSTRACT

Keywords: Many animal species evolved some form of body armor, such as scales of fish and bony plates or osteoderms of Biomimicry reptiles. Although a protective function is often taken for granted, recent studies show that body armor might Osteoderm comprise multiple functionalities and is shaped by trade-offs among these functionalities. Hence, despite the fact Micro-computed tomography that natural body armor might serve as bio-inspiration for the development of artificial protective materials, Reverse-engineering, stress simulation focussing on model systems in which body armor serves a solely protective function might be pivotal. In this study, we investigate the osteoderms of Glyptotherium arizonae, an extinct armadillo-like mammal in which body armor evolved as protection against predators and/or tail club blows of conspecifics. By using a combination of micro-computed tomography, reverse-engineering, stress simulations and mechanical testing of 3D printed models, we show that the combination of dense compact layers and porous lattice core might provide an optimized combination of strength and high energy absorption.

1. Introduction Miocene to Pleistocene, but went extinct some 10,000 years ago. Many glyptodonts were large, up to size of a small car, and possessed a solid Various forms of protective body armor are present in the animal protective carapace consisting of interlocking osteoderms. The carapace kingdom, including the scales of fish and pangolins, the carapaces of of glyptodonts was relatively rigid, unlike the more flexible organiza- turtles and osteoderms - bony plates embedded in the skin of crocodiles tion present in armadillos (Chen et al., 2011). It has been proposed that and armadillos (Yang et al., 2012, 2013). Natural body armor comes in the armor of glyptodonts evolved either to resist attacks by large co- many shapes, ranging from overlapping, highly-flexible elements (Yang occurring predators, or more plausible, serve as protection against the et al., 2012, 2013) to rigid hexagonal structures interlocked with su- powerful tail-blows of conspecifics during intraspecific fights tures (Yang et al., 2015). The diversity in structure and associated (Alexander et al., 1999; Blanco et al., 2009; Arbour and Zanno, 2018). mechanical behavior has made natural body armor a promising can- Unlike reptiles, these mammals did not require thermal regulation, didate for the bio-inspiration of artificial protective materials (Yang eliminating this role for the dermal armor. Yet, despite its (presumably) et al., 2012, 2013; Chintapalli et al., 2014; Naleway et al., 2015; Wang sole protective function, the mechanical behavior of glyptodont osteo- et al., 2016; Torres and Lama, 2017). However, recent studies show that derms has never been investigated. natural body armor might not only fulfil a protective role, but is instead In this study, we examine the mechanical behavior of glyptodont shaped by multiple selective pressures (Broeckhoven et al., 2017). For osteoderms, with particular focus on the structural parameters that example, Broeckhoven et al. (2017) demonstrate a functional trade-off might be crucial to its protective function. To do so, we employ a between strength and thermal capacity of osteoderms in two species of combination of micro-computed tomography techniques, computer si- girdled lizards. Hence, appropriate selection of model systems for the mulation and mechanical testing, using reverse-engineered osteoderms bio-inspiration of artificial body armor, i.e. those that serve a (almost) produced by additive manufacturing. By analyzing the compressive solely protective function, is pivotal for the advancement of protective behavior and energy absorption capabilities of the glyptodont osteo- materials. derms, we aim to provide a better understanding of the configuration Glyptodonts or glyptodontines (Fig. 1) are extinct mammals be- required for carrying out a protective function, which ultimately can longing to the order Cingulata, which includes modern-day armadillos serve to improve the bio-inspiration of artificial body armor. (Delsuc et al., 2016). They roamed the American continent from the

⁎ Corresponding author. E-mail address: [email protected] (A. du Plessis). https://doi.org/10.1016/j.jmbbm.2018.03.037 Received 16 February 2018; Received in revised form 26 March 2018; Accepted 28 March 2018 Available online 29 March 2018 1751-6161/ © 2018 Elsevier Ltd. All rights reserved. A. du Plessis et al. Journal of the Mechanical Behavior of Biomedical Materials 82 (2018) 218–223

properties. This material was selected as it is particularly well suited to additive manufacturing of bio-inspired models. All simulation results were compared relative to one another only, therefore the conclusions made from these simulation results are valid for all materials in the linear elastic regime. However, in this work the testing was conducted with polymer printed models, due to limitations in the maximum compressive force of the available test equipment. Mechanical tests were conducted using a universal test machine (Model 43, MTS, 30 kN maximum force). Compression testing was performed at 1.5 mm per second. Absorption energy was calculated from the obtained stress- strain data according to the procedures in Maskery et al. (2017) and originally in Gibson and Ashby (1999). Of each different design varia- Fig. 1. Left: photograph of a fossil reconstruction of a glyptodont (here: tion, stress-strain curves were obtained for 3 replicates. Panochthus frenzelianus) showing its body armor consisting of interlocking osteoderms. Top right: photograph of an isolated Glyptotherium arizonae osteo- 3. Results and discussion derm used in this study. Bottom right: three-dimensionally rendered micro-CT image of the isolated osteoderm showing the compact layers surrounding the 3.1. Morphological analysis and reverse engineering cancellous core. Osteoderms of Glyptotherium arizonae consist of a cancellous trabe- 2. Methods cular core sandwiched between two compact layers. Each hexagonal osteoderm is joined to the adjacent osteoderms with sutures, thereby 2.1. Micro-computed tomography and reverse engineering of osteoderms making a rigid shield or carapace, as seen in other natural body armor (Chen et al., 2011, 2015; Yang et al., 2015; Achrai and Wagner, 2017). An isolated osteoderm of Glyptotherium arizonae was purchased from A three-dimensionally rendered image of an osteoderm micro-CT scan a commercial dealer (Prehistoric Florida, Tallahassee, FL) and microCT is shown in Fig. 1. Several quantitative measurements were obtained scanned at high resolution using a GE Phoenix v|tome|x L240 dual tube from the micro-CT scan, including the thickness of the compact layer, CT instrument (Phoenix X-ray; General Electric Sensing & Technologies, the mean strut thickness of the cancellous core, and the total porosity of Wunstorf, Germany) located at the Stellenbosch micro-CT facility (Du the cancellous core. An internal rectangular region-of-interest (ROI), Plessis et al., 2016). Following the guidelines in Du Plessis et al. measuring 10 × 10 × 4 mm, was digitally extracted from the micro-CT (2017a), micro-CT scanning was conducted at 120 kV, 100 µA and with scan and was used to characterize the cancellous core in subsequent a voxel size of 30 µm. Data analysis was performed in Volume Graphics analyses (Fig. 2). Total porosity was measured using basic morpholo- VGStudioMax 3.0. Simplified reverse engineered models were created gical tools. Strut thickness measurement was performed using the based on the morphological measurements taken from micro-CT data. maximal-spheres method to find the actual thickness at every point in The models with lattice structures were created in the software Mate- the structure, thereby producing a strut thickness distribution from rialise Magics, including the lattice structure module. Models were which we obtained a mean value (Fig. 2). Total porosity of this internal produced with EOSINT P380 laser sintering system using PA2200 ROI was 66%, whereas the mean strut thickness was 0.25 mm. These polyamide powder. values, in combination with basic dimensional measurements and measurement of the total porosity of the entire osteoderm including the sides, were then used to determine parameters for a reverse-engineered 2.2. Computer simulations and mechanical testing simplified model with surrounding compact layers and lattice core. All measurements including those used for the reserve-engineered model Static load simulation was performed using a direct voxel-based are shown in Table 1. structural mechanics simulation module (see Broeckhoven et al., 2017; Our simplified reverse-engineered model consisted of a hexagonal Broeckhoven and du Plessis, 2017; Du Plessis et al., 2017b, 2018). Si- shape with 1 mm thick compact outer layer and contained a regular mulations were performed assuming homogenous, elastic material strut-based diamond-design lattice with total lattice porosity of 80% properties. The base was kept fixed and a 1 kN load was applied to a and unit cell 1.5 mm (which results in strut thickness 0.31 mm). In this 2 mm diameter circular area on the top surface, using Ti6Al4V material design, the cancellous core has a similar strut thickness, but in total a

Fig. 2. Slice view of the cancellous core of the osteoderm (left) and its strut thickness distribution (right).

219 A. du Plessis et al. Journal of the Mechanical Behavior of Biomedical Materials 82 (2018) 218–223

Table 1 Another possible advantage of a sandwich structure is the ability of Dimensions of the Glyptotherium osteoderm used for micro-CT analysis and of the compact layer to distribute the stress and minimize the effect of the reverse-engineered model. load angle changes. Changing the angle of the load to 45 degrees Osteoderm Model slightly reduced the maximum von Mises stress (Fig. 5b). This can be explained by the fact that the compact layer is absorbing more load at Max width (mm) 28 28 an angle when compared to a perpendicular load. Height (mm) 12 12 Total porosity (%) 56 56 Porosity of foam core (%) 66 80 3.2.2. Strut thickness variation Foam core strut thickness (mm) 0.25 0.31 Investigating changes in the strut thickness of the internal lattice was performed in two ways. In the first method, this was achieved by slightly higher porosity than that of the actual osteoderm. This is due to thickening the struts directly in the lattice unit cell, which also changed the fact that the sides and compact layers of the actual osteoderm the total porosity of the structure (a thicker strut results in total higher contain some degree of porosity, resulting in a total porosity of 56%. In density). The mechanical properties of lattices and open-cell foams are order to obtain the same total porosity, the reverse-engineered model directly related to the total porosity as described by Gibson and Ashby required a more porous core, in combination with its solid sides. A (1999). Therefore as expected, thicker struts distribute stress better: comparison of the actual and reverse-engineered osteoderms is shown since the effective density of the structure is higher, the stress is dis- in Fig. 3. tributed over a larger material fraction resulting in lower stress for thicker struts (Fig. 5c). However, it is expected that as the density 3.2. Stress simulations continues to increase (i.e., overall structure becomes more compact), the failure becomes more catastrophic. Consequently, there should be a Static load simulations (linear elastic isotropic medium assumption) trade-off between strength and strut thickness (or overall density) to on the reverse-engineered model allowed for the assessment of the ef- maximize impact resistance without the risk of catastrophic failure. fects of changes in design parameters on the mechanical properties (here: stress distribution) of the structure. An example hereof is shown 3.2.3. Unit cell size change in Fig. 4, with the red areas highlighting the highest von Mises stress. Here, we varied several design parameters, including strut thickness Changing the strut thickness is also possible without a density and thickness of the compact layers, as well as changes in load condi- change, when changing the unit cell size (in contrast to the above ‘ tions, to unravel the unique properties of these natural sandwich method of strut thickness change with density change). When the unit ’ structures . cell size is increased, the strut thickness increases but the number of struts in the same volume reduces, which leaves the total density un- 3.2.1. Load area size and angle changed. A series of simulations for three unit cell sizes of the internal lattice in the reverse-engineered model were performed. Our results The size of the load application and the angle at which the load was show that unit cell size does not affect the maximum von Mises stress applied, was investigated. Smaller load areas, corresponding to sharper (Fig. 5d). This indicates that more thinner struts are equivalent to less objects, cause increasingly higher stresses for a given force (see Fig. 5a). thicker struts. This is limited, in practice, to the minimum feature size This can be understood as the natural structure evolved to withstand that can be produced by additive manufacturing. The smallest unit cell blunt impact from large objects such as tail-clubs and not as protection size and hence strut thickness that can be manufactured is typically against sharp objects such as teeth. limited to about 0.15 mm, depending on the system used. However, as a

Fig. 3. Image illustrating the similarity in strut thickness between the osteoderm (left) and simpliﬁed reverse-engineered model (right).

220 A. du Plessis et al. Journal of the Mechanical Behavior of Biomedical Materials 82 (2018) 218–223

Fig. 4. Load simulation result showing the maximum von Mises stress distribution in the reverse engineered model after applying a 2 mm diameter load to the top surface. protective material, more thin struts are preferred as the likelihood of 3.2.4. Presence of compact layers catastrophic failure is reduced – if one strut fails there are many that still carry load when using thinner struts. To investigate the eﬀect of the removal of the compact layers, simulations were performed on identical samples with and without compact layers. Removal of the compact layers resulted in an almost three-fold increase in maximum von Mises stress (Fig. 5e), highlighting

Fig. 5. Results of static load simulations on a reverse-engineered model of a Glyptotherium osteoderm. a, the effect of load area size change on the maximum von Mises stress; b, comparison of maximum von Mises stress between perpendicular load and a 45-degree load for 2 mm cross-sectional load area; c, effect of strut thickness change in combination with density change (i.e., 80%, 63% and 43% porosity, respectively); d, effect of unit cell size in combination with strut thickness change (i.e., 0.3 mm, 0.6 mm and 1.0 mm strut thickness, respectively); e, maximum von Mises stress for models with and without side walls; f, effect of varying side wall thickness while keeping the total porosity constant (internal strut thickness changed to compensate for side wall thickness change).

221 A. du Plessis et al. Journal of the Mechanical Behavior of Biomedical Materials 82 (2018) 218–223 the importance of the compact layers, especially dorsally, in dis- tributing the stress upon impact and limiting damage to the structure.

3.2.5. Thickness of compact layers

The compact layers and their thickness appear to have a major in- fluence on the mechanical properties of the structure. To isolate the effect of the compact layer, we varied the thickness thereof and changed the internal lattice porosity accordingly: as the thickness of the compact layer increased, a lower porosity lattice core was incorporated to keep the total porosity of the model reasonably constant. Simulations were run using models with wall thicknesses of 0.1 mm, 0.5 mm, 1 mm and 1.5 mm. In each case the internal lattice was changed to match the total density of the structure, i.e. a thick-wall structure has very high internal porosity and hence lacks internal support, while a very thin wall design has very thick internal lattice struts. The results of our simulations indicate that the reverse engineered structure might be biomechanically optimized – it has a lower stress value at 1 mm compared to thinner or thicker walls (Fig. 5f). On the one hand, thick side walls might cause high stress values locally in the lattice or the upper layer itself where load is applied - due to lack of sufficient internal support. On the other hand, thin side walls might cause higher stress values on the struts themselves: the side walls do not carry any of the load in this case and all the load is carried by the struts. In this geometry, for a 28 mm-diameter and 12 mm height hexagon, the optimal shell thickness is near 1 mm. Fig. 7. Energy absorption for different lattice density (top), and for samples with varying compact layer thickness (bottom). The average value (i.e., mean of 3.3. Mechanical testing and energy absorption the three replicates) ± standard deviations are shown.

Firstly, we tested the effect of variation in lattice density while compact layer has no yield strength and continuously yields up to full keeping the thickness of the compact layer constant. As expected, a densification, whereas the model with compact layers has a higher yield denser sample has a higher yield strength and a steeper stress increase strength and then a plateau region before final densification. The after yielding. As the sample density reduces, the yield stress reduces above-mentioned results indicate that the internal lattice plays an im- (Fig. 6). Secondly, we examined the role of the compact layer sur- portant role in the strength of the structure, with a denser lattice being rounding the cancellous core. The reverse-engineered model without a stronger for both initial yield strength, as well as strength after yielding, for the polymer material used in the tests. The role of the compact layers is therefore shown to be crucial to provide initial yield strength, i.e., it provides protection against initial yielding. Next, we calculated the energy absorption for deformation up to 25% from initial sample height for the compression tests. Our results show that lower porosity, as well as the presence of the compact layer increases the energy absorption (Fig. 7). As the internal lattice becomes denser, the protective role increases based on total energy absorption and strength, but it is expected that its yielding behavior will become more catastrophic (depending on the material ductility). As the shell thickness increases from 0 to 0.5 mm, the energy absorption increases as expected due to the protective role of the shell and the combined effect of the lattice and shell. As the shell increases in thickness to 1 mm, the lattice struts become thinner and more difficult to manu- facture accurately, resulting in reduced energy absorption. As the shell thickness increases to 1.5 mm, at which point it does not contain any internal supports, the energy absorption increases but the failure is catastrophic. This series of experiments demonstrates the compromise between the unique combination of compact layers and lattice core, providing a protective role. It must be noted that in these experiments, a polymer material was used, which is more ductile than bone. In the case of real osteoderms, a lower-density core is required to prevent brittle and catastrophic failure of the bone material. While a polymer material was used for these experiments, this work demonstrates a general biomimetic approach to designing an impact protective geometry using additive manufacturing, which can be applied to any ma- Fig. 6. Top: stress-strain data obtained from compression of reverse-engineered terial. Future work using Ti6Al4V is planned which will further de- samples with varying lattice porosity (i.e., 43%, 62%, 80% porosity of the in- monstrate the methodology. Due to its biomedical relevance, additive ternal lattice, respectively). Bottom: comparison of stress-strain curves for manufacturing of Ti6Al4V has been extensively investigated and opti- samples with and without compact layers. mized to the point where today almost defect-free production of

222 A. du Plessis et al. Journal of the Mechanical Behavior of Biomedical Materials 82 (2018) 218–223 complex geometries are possible. Ceramics are used widely in impact from a structural mechanics viewpoint. Biol. Lett. 13, 20170293. protective devices but current additive manufacturing of ceramics are Broeckhoven, C., du Plessis, A., Hui, C., 2017. Functional trade-off between strength and thermal capacity of dermal armor: insights from girdled lizards. J. Mech. Behav. limited by maximum size, microporosity and other defects, hence this Biomed. Mater. 74, 189–194. might be useful in the future once these limitations have been solved. Chen, I.H., Kiang, J.H., Correa, V., Lopez, M.I., Chen, P.-Y., McKittrick, J., Meyers, M.A., 2011. Armadillo armor: mechanical testing and micro-structural evaluation. J. Mech. Behav. Biomed. Mater. 4, 713–722. 4. Conclusions Chen, I.H., Yang, W., Meyers, M.A., 2015. Leatherback sea turtle shell: a tough and flexible biological design. Acta Biomater. 28, 2–12. Osteoderms of an extinct glyptodont were analyzed using high re- Chintapalli, R.K., Mirkhalaf, M., Dastjerdi, A.K., Barthelat, F., 2014. Fabrication, testing fl fi solution micro-CT, simulations and mechanical testing of reverse-en- and modeling of a new exible armor inspired from natural sh scales and osteoderms. Bioinspiration Biomim. 9, 036005. gineered models with varying morphological parameters. Our results Delsuc, F., Gibb, G.C., Kuch, M., Billet, G., Hautier, L., Southon, J., Rouillard, J.M., show that the combination of dense compact layers and a porous lattice Fernicola, J.C., Vizcaíno, S.F., MacPhee, R.D., Poinar, H.N., 2016. The phylogenetic ffi – core contribute to the strength of the structure and the prevention of a nities of the extinct glyptodonts. Curr. Biol. 26, R155 R156. fi Du Plessis, A., le Roux, S.G., Guelpa, A., 2016. The CT scanner facility at Stellenbosch catastrophic failure. These ndings suggest that the osteoderms of University: an open access X-ray computed tomography laboratory. Nucl. Instrum. Glyptotherium might be biomechanically optimized structures. It is en- Methods Phys. Res. Sect. B Beam Interact. Mater. At. 384, 42–49. visaged that this study assists in the search for optimal protective de- Du Plessis, A., Broeckhoven, C., le Roux, S.G., 2018. Snake fangs: 3D morphological and mechanical analysis by microCT, simulation and physical compression testing. signs to be used in future additively manufactured protective materials. GigaScience 7, 1–8. Du Plessis, A., Broeckhoven, C., Guelpa, A., le Roux, S.G., 2017a. Laboratory X-ray micro- Acknowledgements computed tomography: a user guideline for biological samples. GigaScience 6, 1–11. Du Plessis, A., Yadroitsava, I., le Roux, S.G., Yadroitsev, I., Fieres, J., Reinhart, C., Rossouw, P., 2017b. Prediction of mechanical performance of Ti6Al4V cast alloy Materialise is acknowledged for providing a temporary license of based on microCT-based load simulation. J. Alloy. Compd. 724, 267–274. Magics with the structures module, which was used for model and Gibson, L.J., Ashby, M.F., 1999. CellularSolids: Structure and Properties. Cambridge university press. lattice generation. We thank Johan Els and Andre Heydenrych of the Maskery, I., Aboulkhair, N.T., Aremu, A.O., Tuck, C.J., Ashcroft, I.A., 2017. Compressive Centre for Rapid Prototyping and Manufacturing (CRPM, CUT), for failure modes and energy absorption in additively manufactured double gyroid lat- providing 3D print services for the physical models. We thank Dean tices. Addit. Manuf. 16, 24–29. Kouprianoff for his assistance with mechanical testing. Naleway, S.E., Porter, M.M., McKittrick, J., A Meyers, M., 2015. Structural design elements in biological materials: application to bioinspiration. Adv. Mater. 27, 5455–5476. References Torres, F.G., Lama, D., 2017. Failure retardation in body armor. Bioinspired Biomim. Nanobiomaterials 6, 37–50. Wang, B., Yang, W., Sherman, V.R., Meyers, M.A., 2016. Pangolin armor: overlapping, Achrai, B., Wagner, H.D., 2017. The turtle carapace as an optimized multi-scale biological structure, and mechanical properties of the keratinous scales. Acta Biomater. 41, – – composite armor a review. J. Mech. Behav. Biomed. Mater. 73, 50 67. 60–74. Alexander, R.M., Fraiña, R.A., Vizcaíno, S.F., 1999. Tail blow energy and carapace frac- Yang, W., Chen, I.H., Mckittrick, J., Meyers, M.A., 2012. Flexible dermal armor in nature. – tures in a large glyptodont (Mammalia, Xenarthra). Zool. J. Linn. Soc. 126, 41 49. JOM 64, 475–485. Arbour, V.M., Zanno, L.E., 2018. The evolution of tail weaponization in amniotes. Proc. R. Yang, W., Chen, I.H., Gludovatz, B., Zimmermann, E.A., Ritchie, R.O., Meyers, M.A., Soc. Lond. B Biol. Sci. 285, 20172299. 2013. Natural flexible dermal armor. Adv. Mater. 25, 31–48. Blanco, R.E., Jones, W.W., Rinderknecht, A., 2009. The sweet spot of a biological Yang, W., Naleway, S.E., Porter, M.M., Meyers, M.A., McKittrick, J., 2015. The armored hammer: the centre of percussion of glyptodont (Mammalia: Xenarthra) tail clubs. carapace of the boxfish. Acta Biomater. 23, 1–10. Proc. R. Soc. Lond. B Biol. Sci. 276, 3971–3978. Broeckhoven, C., du Plessis, A., 2017. Has snake fang evolution lost its bite? New insights

223